Pqr Persona Builder

---
name: pqr-persona-builder
description: Build a real ad persona with the PQR method — Demographics, Problems, Questions, Roadblocks, Results — so copy speaks the customer's words instead of marketing-speak. Use when asked to build a persona, define a target audience, do audience research for ads, work out who to target, write a customer profile, find customer pain points, or someone says 'I don't know who my customer is'. For turning the persona into ad angles, see ten-angles-copy. For the full variation matrix, see bulk-ad-variations.
metadata:
  version: 1.0.0
---

# PQR persona builder

Five questions and you have a real person, not a vague guess. Everything downstream —
the angle, the hook, the copy — pulls from this profile, so a weak persona produces a
hundred weak ads.

The rule that makes it work: **use their real words.** "Stay-at-home dads", not
"primary household maintainers" — that's marketing-speak, and the customer scrolls
past words they'd never say.

## Before you start

| Input | Why |
| --- | --- |
| The product, price, and what it actually does | The persona is built relative to a specific offer |
| A rough audience guess | The starting point PQR sharpens |
| Real customer inputs, if any — reviews, DMs, sales calls, comments | The source of their actual words |

If real customer language exists, mine it first. Invented pain points produce ads
that answer doubts nobody has.

## Step 1 — Answer the five PQR questions

| Block | Question | Bad answer | Good answer |
| --- | --- | --- | --- |
| **D**emographics | Who are they? Age, stage, location, income | "25–45 professionals" | "Final-year students and 1–3 yr employees in tier-1/2 India" |
| **P**roblems | What hurts, daily? | "Needs upskilling" | "Applied to 40 jobs, no callbacks; degree feels worthless" |
| **Q**uestions | What do they ask and search at midnight? | "How to grow" | "Why am I not getting interview calls with a first-class degree?" |
| **R**oadblocks | What did they try that failed? | "Lack of skills" | "Did two online certificates; nothing changed" |
| **R**esults | The outcome they are chasing | "Career growth" | "An offer letter in 30–60 days, a skill that pays" |

P, Q, R is the heart. Demographics and Results frame it.

## Step 2 — Run the persona prompt when starting from scratch

The corpus prompt, verbatim structure:

```
You are a customer research expert. I sell [PRODUCT] to [ROUGH AUDIENCE].
Build one detailed audience persona using the PQR method. Give me:
Demographics (age, stage, location, income). Problems (what hurts daily).
Questions (what they ask and search). Roadblocks (what stopped them, what
they tried that failed). Results (the outcome they want).
Use their real words. Be specific, not generic. Write it as a profile I can
paste into my next prompt.
```

Then correct it against anything real you have — reviews and sales calls outrank
the generated draft wherever they disagree.

## Step 3 — Build 2–3 distinct personas, not more

More than three and they blur into each other. Two personas that differ only in age
are one persona — they must differ in **Problems and Roadblocks**, because that is
what changes the ad.

## Step 4 — Map each block to where it gets used

| Persona block | Feeds |
| --- | --- |
| Problems | Callout, Question, Confession angles; the hook's pain line |
| Questions | Secret angle; search-intent headlines |
| Roadblocks | Enemy, Mistake, Compare angles; objection sections |
| Results | Proof, Number, Future angles; the promise in the offer |
| Demographics | Placement and creative casting, not copy |

## Output

```
# Persona: <name them — "Final-year Priya">

Demographics: <age, stage, location, income>
Problems:     <2-4, in their words>
Questions:    <2-4 verbatim searches/asks>
Roadblocks:   <what they tried that failed, what they wrongly blame>
Results:      <the outcome, concrete, with a timeframe if real>

Distinct from persona 2 because: <different Problems/Roadblocks, one line>
Source of language: <reviews / calls / comments / generated-then-corrected>
```

## When it breaks

| What you see | What it means | The fix |
| --- | --- | --- |
| Persona reads like a LinkedIn bio | Demographics-only thinking | P, Q, R is the heart — fill those from real pain |
| Copy from this persona sounds generic | Words are yours, not theirs | Mine reviews, DMs, call notes for verbatim phrases |
| Two personas produce identical ads | They differ by age, not by Problems | Merge them, or find the genuinely different pain |
| Five+ personas on the table | Segmentation theatre | Cut to 2–3; more blur into each other |
| Persona claims a pain the product can't solve | Wishful research | Rebuild Results around what the offer actually delivers |
| No real customer language exists yet | Pre-launch product | Generate with the prompt, label it assumed, replace after 20 real conversations |

Never present a generated persona as researched. Label which blocks came from real
customer language and which are assumptions to verify.

## Rules

- **Use their real words**, because the customer scrolls past vocabulary they would
  never say — "primary household maintainers" loses the stay-at-home dad it names.
- **Differentiate personas by Problems and Roadblocks, never by demographics alone**,
  because only pain changes the ad.
- **2–3 personas maximum**, because beyond that they blur and the matrix explodes.
- **Never invent statistics about the audience** — sizes, incomes, percentages come
  from the operator or stay unstated.
- **Label assumed blocks as assumed**, because a guessed persona presented as
  research misleads every skill downstream.

## Related skills

- **ten-angles-copy** — turns this persona into ten different ad arguments.
- **bulk-ad-variations** — Persona × Angle × Offer at volume; this skill builds the
  first axis.
- **meta-ads-creative** — uses the persona when no reference ad exists.